Figure 4. . Classifying siblings versus unrelated populations, when families with twins and other sibling pairs of the same family were not included in the training set (HCP dataset). (A–F) show the distributions of (A) full accuracy, (B) siblings, (C) unrelated, (D) monozygotic, (E) dizygotic, and (F) nontwin sibling pairs for the connectotyping-based SVM classifier versus the results obtained when the same classifiers were run after randomizing the labels (sibling or unrelated). (G) The consensus’ distribution of ROIs per functional network used in the classifier. (H) The location of such ROIs on the surface of the brain.